If you joined us on March 11 for the webinar from Aricent, you heard valuable insights into how service providers are using software-defined networking (SDN) and network functions virtualization (NFV) for applications such as cloud EPC, virtual CPE, virtual WLAN access controller, and virtual OSS. If you missed it or want to watch again, we’re posting the video here so you can learn how service providers can use SDN and NFV to help reduce costs, increase monetization opportunities, and enhance customer experience.
After the presentation and demo, Aricent’s Saro Velrajan, director of technology, and R. Ezhirpavai, assistant vice president of technology, took questions from demo participants. Read that Q&A below, watch the full presentation
How is it possible to automate configuration or provisioning? Is there a tool for it?
Aricent: SDN offers programmability of the network and network services. For example: A subscriber can sign up for new services from the customer services portal. Service configuration can be translated into network configuration and rolled out into the network in real time without admin involvement. Today, many OEMs provide tools for automating the service provisioning of their equipment.
Are there any standardized tests with regards to conformance testing?
Aricent: We have conformance test suites for OpenFlow v1.3 compliance testing — any NFV instance using OpenFlow can be tested using our compliance test suite. As you may know, NFV standards are still evolving and hence, there is not much in the way of of “standardized tests” for virtualized functions.
Google has shown how using SDN they achieved almost 100 percent link utilization. How can SDN increase the link load without being affected by high delays?
Aricent: Google achieved 100 percent link utilization by carefully prioritizing the critical traffic over the non-critical traffic in the network. With the SDN architecture, the traffic classification and prioritization can be done in real time based on network conditions. For critical traffic, the latency will be almost negligible.
It seems that the assumption of the presentation here is that ISPs will use SDN only. Is that anywhere close to reality?
Aricent: The use cases presented in the webinar highlight greenfield networks. However, SDN and NFV applications can be rolled out in hybrid networks also. Service providers will migrate to a hybrid architecture before completely moving to SDN-ized architecture.
The CPE can contain very useful diagnostic capabilities. Should we expect many of those to continue to reside in a physical CPE?
Aricent: Application-specific statistics and counters can be retrieved by service providers to offer differentiated/value-added services to subscribers. Though debugging/diagnostic information will be retrieved periodically by the northbound systems, select debugging/diagnostics, such as debugging a kernel crash, can be done only in the physical CPE.
Would it be a different story if the CPEs are in subscribers’ networks?
Aricent: Yes, the workflows will differ for enterprise CPEs versus residential CPEs.
Do you expand the OpenFlow spec to support access point management?
Aricent: OF-Config or Capwap can be used for AP management. OpenFlow is used for flow-table management.
Any good examples today of OEMs or carriers using/deploying virtual OSS solutions?
Aricent: Several Tier-1 OEMs have already started developing virtualized OSS solutions. However, we have yet to see any active deployments.
What is TR.069 used for in Slide 33? Isn’t TR.069 an ancient protocol replaced by Netconf and others nowadays?
Aricent: Yes, TR.069 is ancient. However, we still see TR.069 used in networks, and Aricent’s vCPE solution can be integrated with a third-party TR.069 stack.
If the CPE is virtualized, what is the communication mechanism between what is left of the CPE and the cloud?
Aricent: OpenFlow can be used for flow management, and OF-Config can be used for device management.
What are the possible solutions for VoIP latency problems in SDN-based networks offered by Aricent?
Aricent: We have a dynamic bandwidth management SDN application, which provides application-aware QoS profiles configuration. It helps to address the network latency.
Is the carrier/operator interest in BRAS virtualization meant to be as an overlay to existing BRAS or a replacement on a per-region or per-geography basis? That is, how would you introduce that into legacy environments?
Aricent: In the near term, it is going to follow an overlay architecture adding to the existing capacity. We see operators doing early trials and evaluations of the virtual BRAS solution. Several subscriber-facing services can be gradually virtualized to minimize the complexity involved in integrating with legacy environments. For example, functions such as subscriber policy management and route propagation can be easily done without impacting the current architecture significantly.